Earnings management, firm location, and financial reporting choice

Earnings management, firm location, and financial reporting choice: An analysis of fair value reporting for investment

properties in an emerging market

Presented by

Dr Kin Lo

Associate Professor University of British Columbia

#2013/14-02

The views and opinions expressed in this working paper are those of the author(s) and not necessarily those of the School of Accountancy, Singapore Management University.

1

Earnings management, firm location, and financial reporting choice: An analysis of fair value reporting for investment properties in an emerging market

Chen Chen [email protected]

School of Management China University of Mining and Technology

Sanhuan South Road, Xuzhou, Jiangsu, China, 221116

Kin Lo [email protected]

Sauder School of Business University of British Columbia

2053 Main Mall, Vancouver, British Columbia, Canada, V6T 1Z2

Desmond Tsang [email protected]

Desautels Faculty of Management McGill University

1001 Sherbrooke Street West, Montreal, Quebec, Canada, H3A 1G5

Jing Zhang [email protected]

Desautels Faculty of Management McGill University

1001 Sherbrooke Street West, Montreal, Quebec, Canada, H3A 1G5

April, 2013

We thank Jeff Callen, Steve Fortin, Jingjing Zhang, Zvi Singer, seminar participants at McGill University Research Workshop, and anonymous reviewers of the American Accounting Association Annual Meeting 2012 and the Canadian Academic Accounting Association Conference 2012 for their helpful comments. Tsang and Zhang acknowledge financial support from McGill University and the Canadian Academic Accounting Association.

2

Earnings management, firm location, and financial reporting choice: An analysis of fair value reporting for investment properties in an emerging market

Abstract: In this study, we examine firms’ decision to adopt fair value accounting for investment properties, and how firm and property locations can affect this financial reporting choice. Unlike financial assets reported at fair value, investment properties are unique and cannot be traded on an exchange. Hence, fair value estimation on investment properties is less verifiable and can be subject to more managerial discretion. As investment properties are location-specific, firms also have more opportunities to misstate fair values where the real estate markets are illiquid and investors’ monitoring is low. Utilizing the emerging market setting of China, we find evidence that the fair value option for investment properties is more likely to be chosen by firms that had significant prior earnings management activities. We also find that earnings management firms are more likely to adopt the fair value model when the firms’ headquarters and investment properties are located in less developed regions. Confirming that firm chooses the fair value model to manage earnings, we show that firms choosing the fair value model use unrealized gains and losses associated with investment properties to smooth earnings or to beat earnings benchmarks. Overall, our findings indicate fair value reporting decision for investment properties in the emerging Chinese market is primarily driven by managerial opportunism. Keywords: fair value reporting; earnings management; firm location; investment properties; emerging market

3

1. Introduction For accounting purposes, investment properties are real estate held for the purpose of earning

rents or capital appreciation. By its nature, real estate holdings are location specific,

differentiated, and unique. As a result, they cannot be traded on an exchange in contrast to

financial assets. Hence, fair value estimation on investment properties is likely to be less

verifiable and can be subject to more managerial manipulation. While prior studies have

generally established the relevance of fair value reporting (Barth and Clinch 1998; Sloan 1999),

researchers have also shown fair values of investment properties are used for earnings

management purposes (Dietrich et al. 2001). With the adoption of the International Financial

Reporting Standards (IFRS) in Europe on January 1st of 2005, publicly traded European firms are

now provided with fair value guidance on investment fixed assets and are allowed the option of

reporting investment properties at fair values under International Accounting Standards 40 (IAS

40). Recent studies document that, subsequent to the adoption of IAS 40, most firms choose to

report fair values for their investment properties in order to reduce agency costs and information

asymmetry (Muller et al. 2011; Edelstein et al. 2012). Moreover, the choice of adopting the fair

value model appears to be negatively related to managerial opportunism (Quagli and Avallone

2010). Hence, the extant literature seems to indicate that IAS 40 has improved the transparency

and quality of reporting with respect to investment properties.

In this study, we examine how fair value reporting for investment properties is applied in

an emerging market setting. Ball et al. (2003) shows financial reporting quality is more strongly

influenced by local preparers’ incentives than by the quality of the accounting standards. As

more and more countries start to adopt IFRS or allow their publicly listed firms to report IFRS-

equivalent standards, researchers have raised questions as to whether IFRS can be effectively

applied in the same way across different countries (Leuz et al. 2003). IFRS are commonly

considered as principle-based standards that require substantial professional judgment and

regulatory guidance. Thus, effective implementation of IFRS requires a well-governed and well-

developed financial reporting environment. In the context of investment properties, the adoption

of IFRS is especially complicated because fair value reporting also requires reliable estimates

from a reputable appraisal industry with high quality governance and regulation. Moreover, the

quality of fair value estimates depends on a real estate market that is efficient and liquid.

However, real estate markets in developing economies are typically less efficient and real estate

4

transactions in these countries tend to have lower transparency.1 Hence, it is questionable

whether firms in emerging markets would embrace fair value reporting for investment properties,

and if so, to what extent managers would provide impartial and reliable fair value estimates for

these properties.

The emerging market we choose to examine is China. Since 2007, the Chinese

Accounting Standards Board has developed the Chinese Accounting Standards 3 (CAS 3) on the

accounting for investment properties. The standards are very similar to IAS 40, giving firms the

option of reporting on the balance sheet their investment properties at fair values instead of

historical costs.2 We analyze all publicly traded Chinese firms with investment properties over

the period of 2007 to 2009, and document that very few firms (3.8%) in China chose the fair

value model to account for their investment properties. This is a marked difference from

evidence documented in developed markets.3 This low adoption rate is consistent with a recent

report published by the Chinese CA Network (2011), and it could be the result that fair value

estimation is just too costly in relation to the benefit of financial statement users (IASC 1999,

paragraph B46).4 However, the lukewarm reception can also be concurrent with anecdotal

comments of the Chinese financial press (e.g., Sohu Business 2008), that fair value reporting is

not easily implemented and more subject to manipulation with the lack of an efficient real estate

market and a reputed, experienced appraisal industry in China.5

We investigate the type of firms that choose to report investment properties at fair values,

and find that they are predominantly those that had a history of earnings management activities

(measured by firms’ discretionary accruals in the past five years). Our findings differ

significantly from prior IFRS studies, which show fair value firms under IAS 40 are committed

to greater reporting transparency (Muller et al. 2011), less managerial opportunism (Quagli and

1 For example, see the Global Real Estate Transparency Index 2010 Report by Jones Lang LaSalle. 2 However, unlike IAS 40, firms that choose to report historical costs under CAS 3 are not required to disclose investment properties’ fair values in the notes to the financial statements. 3 For instance, Edelstein et al. (2012) show that 75% of firms in their sample adopt the fair value model to account for investment properties. 4 This is especially true in an emerging market where it is more difficult to obtain reliable fair value appraisals. 5 One may argue that the tax factor plays a role in the unenthusiatic response to the adoption of fair value accounting for investment properties in China, especially given that the Chinese real estate market has experienced unprecedented growth in the last decade. However, the tax policy in China forbids the inclusion of unrealized fair value gains and losses of investment properties in taxable income, thus ruling out tax being the contributing factor to the unpopularity of the fair value accounting rule.

5

Avallone 2010), enhanced comparability (Cairns et al. 2011) and higher disclosure standards

(Edelstein et al. 2012).

Within the Chinese market, we further examine the impact of firm location on the

relationship between the use of fair value and past earnings management. Location is important

for several reasons. First, as better developed regions have more active and efficient real estate

markets, yielding external indications of fair value that are more frequent and more reliable.

Second, the real estate appraisal industry is likely to be more mature with more qualified experts

to conduct appraisals. Moreover, under the knowledge spillover theory, information asymmetry

is lower when average education level of people is higher in more developed regions.

Consequently, investors’ monitoring is also stronger in developed regions. Thus, the location of

firms and more specifically their investment properties have significant bearing on the fair value

reporting decision because it determines whether and to what extent there are opportunities to

manage earnings.

Based on results from a cross-country setting, prior studies (e.g., Muller et al. 2011)

suggest that fair value could be more prevalent in developed regions as managers strive to

improve transparency. On the other hand, the managerial opportunism explanation suggests that

fair value would be more commonly used in less developed regions. Our findings are consistent

with the latter hypothesis, indicating that the fair value option for investment properties tend to

be used as an earnings management tool.

Having established that the past history of earnings management and the opportunity to

bias fair value estimates are important factors in managers’ choice of fair value accounting, we

close the analysis by examining the outcomes of those accounting choices. Lo (2008) discusses

that managers who manage earnings must have some motives to mislead investors or to alter

contractual outcome. In this study, we examine whether firms use their fair value reporting

decision to achieve two earnings management goals commonly documented in prior literature:

To smooth earnings (Trueman and Titman 1988) and to meet earnings benchmark (Burgstahler

and Dichev 1997; DeGeorge et al. 1999). We find that fair value firms use unrealized fair value

gains and losses to reduce earnings volatility, and this smoothing behavior is again more

apparent in less developed regions. Furthermore, we also find that fair value adopting firms are

more likely to meet or beat certain earnings benchmarks. The findings confirm our hypothesis

that these firms are not adopting the fair value model to provide more transparent and value

6

relevant information to investors. Instead, firms choose to report fair values for investment

properties for manipulative purposes to make use of the flexibility allowed in fair value

estimation.

Collectively, our results imply that fair value reporting for investment properties in China

has resulted in outcomes that are drastically different from those documented in developed

economies. It seems that firms tend to choose fair value reporting primarily to manage earnings.

The fair value option is particularly attractive for firms located in less developed regions, where

there exists the greatest opportunity to bias fair value estimates. These findings support the

argument made by prior studies (e.g., Leuz and Wysocki 2008; Ball et al. 2003) that high quality

accounting standards cannot be effective without the support of a well-developed financial

reporting environment.

We choose the emerging Chinese market to conduct our analysis for several reasons. First,

CAS 3 is drafted based on IAS 40, allowing us to evaluate the effect of similar fair value

regulation in an emerging market context. At present, other large developing countries (for

instance, India) have not developed standards equivalent to IAS 40. The single country setting

also allows us to isolate the effect of location from other legal, institutional, tax and cultural

factors. Second, the financial reporting environment in China is in its early stage of development

and many publicly traded Chinese firms, especially those with government connections, are

subject to relatively minimal scrutiny by regulators. Hence, it engenders a culture where earnings

management activities are still considered common and prevalent (Li et al. 2008). Third, China

has some of the largest real estate markets in the world. While some markets are still in their

infancy, other markets (e.g., Shanghai, Beijing) are very developed and active, with highly

transparent property listing system comparable to developed countries. The development

disparity across different regional real estate markets in China provides us a favorable

environment to assess the importance of real estate market liquidity and efficiency on the

accounting choice for investment properties.

Our study contributes to extant literature in two important ways. First, we offer new

evidence that addresses the concerns among academics and industry practitioners on the

reliability of fair value reporting for nonfinancial assets. In particular, we show that, in an

emerging market, the adoption of high quality standards such as IFRS or IFRS-like standards

does not necessarily lead to improvement of financial reporting quality. In fact, the effectiveness

7

of IFRS is dependent on the adopting country’s local environmental factors. We show that fair

value accounting is being implemented perhaps too soon in the current Chinese market.

Consequently, fair value accounting has been used as an earnings management tool. Our results

bear important policy implications to standard setters and regulators governing emerging markets,

and suggest that regulators must carefully consider their IFRS implementation plans as to when

and how to converge local accounting standards to IFRS. Our study also suggests that IFRS may

not be the ‘one size fits all’ standards suitable for all environments.

Our second contribution is from providing evidence that (within-country) location is an

important factor affecting the use of fair value and earnings management. Fields et al. (2001)

discuss three important factors driving accounting choices (i.e., agency cost, information

asymmetry, and externalities affecting non-contracting parties). Firm and investment locations

can affect both agency costs and information asymmetry through investor monitoring. For

investment properties, the influence of location is especially palpable as efficiency of the local

real estate market can affect how these assets are measured on the financial statements. Yet, the

location factor has largely been overlooked by accounting researchers. Among the few existing

studies, Urcan (2007) examines the relationship of firm location and financial reporting quality,

showing that rural firms in the U.S. have higher quality than urban firms. Other studies that link

firm location to capital market research have examined the importance of location on cost of

capital, information asymmetry or market performance (e.g., John et al. 2008; Loughran and

Schultz 2005; 2006; Francis et al. 2008). In broader terms, our research adds to this line of scarce

but important literature by showing the effect of location on financial reporting decision making.

(Among these four studies just cited, only Loughran and Schultz 2005 has been published.)

The rest of the paper is organized as follows. The next section provides some background

information on the development of fair value accounting in China and discusses related literature.

Section 3 develops the hypotheses and outlines our research design. Section 4 describes the

sample selection process and sample statistics. Section 5 reports the empirical results. We

discuss additional and robustness analysis in Section 6 and offer concluding remarks in the last

section.

8

2. Fair value accounting in China and related literature review

2.1. The development of fair value accounting in China The concept of fair values was first introduced in the Chinese Accounting Standards in 1998 for

debt restructuring expenses (CAS 12). However, the Ministry of Finance (MOF) issued revisions

of the standards in 2001 disallowing the fair value model, raising concerns about the objectivity

of fair values due to low market efficiency and the prevalence of earnings management in China.

As the Chinese market witnessed significant improvement in recent years, the MOF decided to

combine the CAS with IFRS-equivalent standards and fair value reporting was reintroduced in

China in 2007.

Under the new CAS, effective January 1st, 2007, fair value reporting applies to

investment properties, debt restructuring, financial instruments and certain non-monetary

transactions. However, firms are allowed the option of choosing the fair value model versus the

historical cost model only for investment properties (CAS 3). It is also the first time investment

properties are defined in the CAS, and the China Securities Regulatory Commission (CASC)

proclaims that all buildings and land held for rental or for capital appreciation

must be reclassified as investment properties. Firms that choose to report their investment

properties under the fair value model would show investment properties on the balance sheet at

fair value and changes in fair value would flow through income. Under the historical cost model,

investment properties appear on the balance sheet at depreciated cost and depreciation expense

flows through income. Compared to IAS 40, CAS 3 has stricter requirements in terms of

application. In particular, CAS 3 does not allow a firm to choose the fair value model unless it

can justify: (a) its investment properties are located in an active market, or (b) similar investment

properties’ market information can be obtained. Moreover, CAS 3 states that a firm should only

choose one model to value all its investment properties, and no subsequent conversion from fair

value model to cost model is allowed. This setting of CAS 3 reflects regulators’ concerns about

the reliability of fair value estimates for investment properties in China.

2.2. Related literature review

2.2.1. Fair value for investment properties On January 1st, 2005, the European Union (EU) required all publicly traded firms to adopt the

International Financial Reporting Standards. The new standards have drastically changed the

9

financial reporting environment of many European firms. One of the more notable changes is the

fair value reporting of nonfinancial assets. IAS 16 allows revaluation of property, plant, and

equipment, and IAS 38 likewise provides this revaluation alternative for intangible assets. For

investment property, IAS 40 provides an option for firms to use either the cost model or the fair

value model. Prior academic studies have looked into the rationales and consequence of firms’

choice in choosing the fair value model. Muller et al. (2011) show firms that have higher

information asymmetry, measured by bid-ask spreads, are more likely to use the fair value model

to report investment properties. They argue it is investors’ demand for transparency that drives

the choice of fair value reporting. Christensen and Nikolaev (2009) associate the choice of fair

value reporting for nonfinancial assets with debt contracting theory. They show that fair value

reporting firms rely more on debt financing than companies using historical costs. They interpret

the findings as indicating that fair value reporting reduces agency costs, by revealing asset exit

values to creditors. Quagli and Avallone (2010) extend the above-mentioned studies and

examine a comprehensive set of factors driving firms’ fair value reporting choice under IAS 40.

Their results indicate that firm size, a proxy for political cost, is the most significant factor of the

fair value choice. On the contrary, they find that leverage and information asymmetry do not

play an important role.6 Cairns et al. (2011) investigate the use of fair values for various assets

(e.g., financial instruments, property, plant and equipment, investment properties) in the U.K.

and Australia under IFRS. They document most companies that hold investment properties adopt

the fair value model. Edelstein et al. (2012) analyze the financial statement disclosure for fair

value firms, and find that these firms make extensive disclosures with regard to their investment

properties according to the requirement of IAS 40.

2.2.2. Reliability of fair values The reliability of fair values has long been a concern for academics and regulators, and most

research focused on financial assets in the U.S. banking industry. Though fair values are

commonly regarded as more value relevant (e.g., Barth 1994; Nelson 1996), other studies also

show that fair values are more subjective and can be manipulated easily for earnings

management purpose. For example, Hitz (2007) postulates that the manipulation of fair values

6 The difference of results could be due to sampling differences, as Quagli and Avallone (2010) examine their hypothesis using a small sample of 73 observations from a subset of real estate companies in seven E.U. countries (Finland, France, Germany, Greece, Italy, Spain and Sweden).

10

can actually hurt investors despite the relevance of fair values. In a theoretical model, this paper

shows that fair value accounting can become exclusively an earnings management tool for risky

companies. Song et al. (2010) examined SFAS 157 classifications of fair value assets into three

levels based on the degree of certainty on the assets’ underlying values. As expected, the value

relevance of the firms’ level 3 assets (i.e., assets that lack reliable market values or valuations

and require significant management assumptions) is significantly lower than their level 2 and

level 1 assets, highlighting the subjectivity and unreliability of estimates based on management

discretion.

The issue of reliability, more recently relabeled as representational faithfulness, is a more

serious concern for nonfinancial assets. Before IFRS, fair value accounting (or revaluation) for

investment properties was only allowed in a few countries (e.g., U.K., Australia). Academic

studies that examine the reliability issues in these markets generally find that fair value is value

relevant (Easton et al. 1993; Aboody et al. 1999). However, value relevance is a low hurdle that

simply measures whether fair values are significantly associated with the market value of equity

(i.e., sign not magnitude).7 Danbolt and Rees (2008) find that fair values become less value

relevant when they are subject to increased managerial discretion. Another study by Dietrich et

al. (2001) shows that fair value appraisals of investment properties are reliable estimates for the

properties’ selling prices, but these estimates require managerial discretion and managers select,

within a reasonable range, the estimates that help the company to report increased earnings or

smooth earning.

2.2.3. Effect of location on financing, disclosure, and investment choices The finance literature contains a few studies showing that firms’ geographical location plays a

role in the capital markets. Loughran and Schultz (2006) compare financing decisions of rural

and urban firms and find that rural firms are less likely to rely on external equity financing. They

show that this is due to higher cost for investors to ascertain the financial performance of rural

firms, which translate to an adverse selection outcome and consequently higher cost of equity

financing for rural firms. Loughran and Schultz (2005) further show trading volume is

substantially higher for firms located in big metropolitan areas. Francis et al. (2008) document

7 Another concern about measuring value relevance using the market value equity is the inherent circularity. If investors already know the value of assets to price the shares, what is the purpose of reporting those values to investors.

11

that firms in remote rural areas exhibit significantly higher cost of debt capital compared with

those in urban areas. Cai and Tian (2009) show firms with urban headquarters are more

transparent with a higher chance of takeover. John et al. (2008) find evidence that firms in rural

areas offer higher dividend yields than those in central cities, as managers in remote firms are

presumed by investors to have more discretion in managing cash flow.

Another theory that predicts a difference between urban and rural firms is related to

knowledge spillover. The knowledge spillover theory postulates that firms in big cities perform

better because managers have more opportunities to network and to build valuable relationships

with their peers. Residents are generally more educated in big cities and are more able to learn

from their peers. Christoffersen and Sarkissian (2009) find a positive relationship between city

size and mutual fund performance, supporting the argument that managers in larger cities are

more experienced and more knowledgeable.8

Aside from the urban/rural differences, geographic distance also matters. Proximity

between firms and their investors alleviates agency cost concerns by facilitating (literally) closer

monitoring. Recent studies on institutional investors find evidences that increases in local

institutional investors predict future stock returns (Baik et al. 2010) and geographical proximity

between firms and institutional investors reduces managers’ opportunistic behaviors and

improves earnings predictability (Ayers et al. 2011).

The effect of geographic proximity is related to the home-bias literature. Prior research

shows fund managers and institutional investors have a home-bias that is not only international

but also regional. This bias is rational to the extent that geographic proximity between firms and

investors confers an information advantage and these investors are able to exploit their specific

knowledge and earn abnormal returns (Coval and Moskowitz 2001; Ivkovic and Weisbenner

2005).9 Finally, the impact of geographical location on financial reporting decisions is an almost

uncharted area in extant literature. To the best of our knowledge, there has been one

(unpublished) study by Urcan (2007), which shows earnings of rural firms in the U.S. have

higher persistence and greater conservatism. The author interprets the findings as being 8 In the urban economics literature, research has also shown a positive relationship between wages, productivity, and the education level of inhabitants in a city (e.g., Rauch 1993; Glaeser et al. 1995; Glaeser and Mare 2001). Some studies also show locating in urban areas can improve technological spillovers (e.g., Audretsch and Feldman 2004; Fritsch 2003). 9 Other studies focus on venture capitalists’ investment strategies and find venture capital managers are also most likely to invest in local firms (e.g., Gupta and Sapienza 1992; Norton and Tenenbaum 1993; Cumming 2006; Sorenson and Stuart 2001).

12

consistent with managers of rural firms attempting to mitigate the adverse effect of location by

providing higher quality reporting.

3. Hypothesis development and research design

3.1. Managerial opportunism There is a long line of literature that shows managerial opportunism as a critical motivation for

accounting choice. Based on the positive theory of accounting (Watts and Zimmerman 1978),

contractual arrangements such as management compensation contracts and bond covenants

influence managers’ financial reporting decisions. Managers make financial reporting decisions

to maximize their compensation (e.g., Healy 1985; Holthausen et al. 1995; Guidry et al. 1999),

or to avoid debt covenant violations (e.g., Defond and Jiambalvo 1994). Research also shows

that managers make opportunistic reporting decisions for capital market reasons (e.g., Kasznik

1999; Perry and Williams 1994; Graham et al. 2005).

In this study, we examine the potential for managerial opportunism in financial reporting

choices relating to investment properties in an emerging market. We examine the accounting

choices for investment properties because manipulation of fair values is perhaps the most critical

concern for the application of fair value reporting on nonfinancial assets such as investment

properties. Compared to financial markets, real property markets tend to be less liquid. As each

property is unique, there are no organized exchanges on which to trade property assets. Hence,

when the E.U. adopted IFRS in 2005 that allow the fair value option for investment properties,

many academics and regulators were and continue to be skeptical because fair value estimates

for investment properties are less verifiable and managers can significantly manipulate such

estimates.10

In emerging economies, the reliability of fair values is an even more serious concern.

First, these economies usually have lower transparency, less developed governance and legal

structure, and higher information asymmetry between corporate insiders and outside investors.

The result is that capital markets are less informationally efficient, so even fair values of

financial assets are of questionable reliability. Second, the determination of fair values for

investment properties relies on appraisals of real estate experts due to the absence of quoted

10 For instance, Herrmann and Sudagaram (2006) suggest that verifiability is the most important reason that U.S. is not following the IFRS on this financial reporting issue and still favor historical costs over fair values in the accounting of nonfinancial assets.

13

prices on exchanges. Given that the real estate market in emerging economies is generally less

liquid and less transparent, the quality of these real estate appraisals is lower (Chinese CA

Network 2011). A third factor is the absence of a property tax system in many emerging

economies. In most developed countries, the property tax base is the fair value of the real estate.

In such systems, government assessors determine the fair value of properties annually,

extrapolating from actual property sales and other data, thereby providing an independent source

of fair value information.11

In the emerging Chinese market that we choose to conduct our analysis, the capital

markets are also characterized by significant insider trading activity and few institutional

investors. Moreover, many publicly traded firms are state-owned enterprises with direct or

indirect connections to the government. In such markets, there is weak oversight of financial

reporting by capital market regulators. Hence, earnings management is more prevalent in China.

Moreover, real estate transactions in China are not very transparent (Wang and Wang 2012). We

posit that, if managers adopt the fair value model to facilitate earnings management, 12 then

firms that have displayed a higher proclivity for earnings management in the past are more likely

to adopt the fair value model. Alternatively, prior research (e.g., Muller et al. 2011; Quagli and

Avallone 2010; Edelstein et al. 2012) shows that the adoption of fair value accounting for

investment properties under IFRS improves financial reporting quality in developed countries. If

the objective of Chinese managers who adopt the fair value model is similar to managers in

developed countries, that is to provide more value relevant information to investors, then the

accounting choice for investment properties should not be related to managerial opportunism.

Hence, we would observe no relationship between the adoption of the fair value model and past

earnings management activity. We present our first hypothesis as follows:

H1: The likelihood of reporting fair values for investment properties in an emerging market is

positively associated with managers’ history of earnings management.

11 On the contrary, for instance, the Chinese property tax system being piloted in two cities is not based on fair values. Instead, it is based on the original price minus 10 to 30 percent of depreciation at a rate of 1.2 percent or levied at 15 percent of the actual rental income of the property (Man 2012). 12 Though firms can also manage earnings by manipulating depreciation accruals in the cost model, Marquardt and Wiedman (2004) show that these accruals are not the primary accounts being managed.

14

We measure past earnings management activity in the years preceding the implementation of the

new Chinese Accounting Policy in 2007. We use the discretionary accruals model from Dechow

and Dichev (2002) to measure earnings management.13 This model maps accounting accruals

into operating cash flows in contemporaneous and adjacent periods, and has been documented to

have the best fraud detection power (Jones et al. 2010). The model is as follows:

ACCRit = b0 + b1CFOi,t-1 + b2CFOi,t + b3CFOi,t+1 + b4∆REVit + b5PPEit + eit (1)

ACCRit is total accruals for firm i in year t, calculated as earnings less operating cash flow.

CFOi,t represents operating cash flows for firm i in year t. ∆REVit is the difference in total

revenues for firm i between year t and t-1. PPEit is the gross value of property, plant and

equipment at the end of year t. All variables are scaled by total assets at the end of year t. The

model is estimated separately for each industry in which there are at least 10 firms. Discretionary

accruals are calculated as the residuals from equation (1). For each firm, we calculate the mean

value of absolute discretionary accruals over the five year period of 2001-2005. The square root

of the average discretionary accruals, labeled as EM, is of the proxy for prior earnings

management.14

EM is the key variable in the test of H1, which we implement by estimating the following

logistic model:

𝐹𝑉it = 𝑏0 + 𝑏1𝐸𝑀i+ 𝑏2𝑆𝐼𝑍𝐸!" + 𝑏3𝐿𝐸𝑉!" + 𝑏4𝐶𝐹𝑂!" + 𝑏5𝑃𝑃𝐸!" + 𝑏6𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" +𝑏!𝐵𝐼𝐺4!" + 𝑏!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" + 𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!!𝐷𝑂𝑀!" + 𝑒!" (2)

The dependent variable, FVit, is an indicator variable that equals to 1 if firm i uses fair values to

report their investment properties at year t, and 0 otherwise. If the choice of the fair value model

is driven by managerial opportunism, we expect b1 to be significantly positive.

We follow prior studies (e.g., Muller et al. 2011; Quagli and Avallone 2010) and include

the following control variables: SIZEit, measured as the log of firm i’s equity market

capitalization, is a proxy of political costs; LEVit, measured as firm i’s total liability divided by 13 As robustness check, we also calculate discretionary accruals using alternative measures of discretionary accruals and obtain similar empirical findings. We discuss the results in a subsequent section. 14 We use square root to transform the average discretionary accruals as its distribution is right-skewed. However, results are qualitatively similar when we use the average discretionary accruals without transformation to conduct our analysis.

15

equity market capitalization, is a control for the leverage effect; CFOit, measured as firm i’s cash

flows from operations divided by total assets; 𝑅𝐸𝑇𝑈𝑅𝑁!", measured as firms i’s annual stock

return; 𝐿𝑂𝑆𝑆!", an indicator variable equal to 1 if firm i had negative net income in the prior year

(0 otherwise); CFOit, 𝑅𝐸𝑇𝑈𝑅𝑁!" and 𝐿𝑂𝑆𝑆!" serve as controls for firm performance; PPEit,

measured as firm i’s property, plant and equipment divided by total assets, is included to proxy

for the magnitude of other fixed assets; Big4it, an indicator variable equal to 1 if firm i is audited

by Big4 auditors (0 otherwise), is a control for audit quality. In addition, we include corporate

governance factors as they have the potential to affect firms’ earnings management and future

financial reporting discretion. We use three governance variables: managerial shareholding

(𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" ), CEO duality (𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" ), and largest five dominant shareholdings

(𝐷𝑂𝑀!" ). Detailed definitions of all variables are included in the data appendix. Finally, we also

include year and industry indicators.

3.2. Effect of firm location Next, we examine the effect of firm location on the association between fair value choice and

earnings management history. Location is an important factor in the estimation of investment

property values because of differences in liquidity of the real estate markets and the quality of

appraisals. Fair value estimates for investment properties should be less subjective if there is an

active and transparent real estate market where auditors (and to some extent, investors) can

easily verify the information. Secondly, as noted earlier, geographic proximity between a firm

and its owners alleviates the conflicts created by the separation of ownership and control.

Furthermore, prior research acknowledges an information advantage (e.g., Ivkovic and

Weisbenner 2005) and knowledge spillover effects (e.g., Christoffsen and Sarkissian 2009) in

larger, urban cities.

Aggregating the above arguments, we believe that fair value estimates for investment

properties in less developed regions are more likely to be subject to managerial opportunism.

Thus, we posit that firms located in less developed regions will show a stronger association

between the choice of fair value model for investment properties and past earnings management

activity:

16

H2: Firms with property located in less developed regions have a stronger association between

the choice of fair value accounting for investment property and their earnings management

history.

Since the disclosure of investment property location is voluntary, we are unable to find the exact

location of investment properties for every firm. Hence, we use the firms’ headquarters to serve

as a proxy for investment property locations, as Chinese firms tend to concentrate on their local

markets and own investment properties close to the firms’ headquarters.15

In prior U.S. literature (e.g., Loughran and Schulz 2006), researchers define urban (or

better developed) regions as the largest 10 metropolitan areas in the U.S. according to the Census,

and rural (or less developed) regions as areas that are 100 miles away from any of the 49 largest

metropolitan areas in the U.S. We cannot follow the same methodology because such

classification assumes economic development is correlated with population of a region. And, in

the case of China, there is a large population in almost every province and city. Hence,

considering the distinct Chinese setting, we elect to construct our own measures of urban

development.

A recent survey by the Shanghai and Shenzhen stock exchanges showed that almost 50%

of investors are concentrated in five regions/cities: Shanghai, Guangdong, Beijing, Zhejiang and

Jiangsu, and the other 50% of investors are dispersed across the remaining 26 provinces (Tao

2008). Moreover, in 2006, the Institute of Social Science of the People's Republic of China

published a blue book of Chinese regional development. It clearly shows the most economically

developed cities are concentrated in the Yangtze River Delta and the Pearl River Delta, followed

by the Beijing-Tian Jing District. These three regions also encompass the above-mentioned five

regions/cities with the most investors. Hence, we construct our first proxy of location, HQ1, an

indicator variable equal to 1 (0 otherwise) if a firm’s headquarter is located in the three most

developed economic areas in China: Yangtze River Delta, Pearl River Delta, and Beijing-Tian

Jing District.

15 We verify this claim by examining a hand-collected sample of firms that voluntarily disclose investment property locations. We have a total of 448 firm-year observations with investment property location information. We find that 74% (329 out of 448) of the observations report investment properties predominantly (i.e., greater than 75% of total investment properties) located in the same city as the firms’ headquarters. We discuss the results of further analysis on this subsample in a subsequent section.

17

We also construct a second measure of location. We create an index for each province

based on seven economic factors: (1) per capita GDP; (2) percentage of stock trading volume for

firms in the province relative to the whole country; (3) geographic distance of the province to the

closer of the two Chinese financial exchanges (i.e., Shanghai and Shenzhen); (4) level of

residential consumption level; (5) percentage of urban population relative to the country; (6)

number of financial experts as a percentage of population; (7) number of real estate experts as a

percentage of population. We separately rank the 35 provinces by each of these seven factors,

and calculate the mean value of each factor as the cutoff.16 The index is constructed for each

province by assigning 1 point for each of the province’s economic factor that is above the cutoff

(0 otherwise). Hence, the most developed province can report a maximum index value of 7. The

index serves as our second proxy of location, HQ2.

We examine the prediction of H2 that the impact of managerial opportunism on the

choice of fair value model for investment properties is more significant for firms located in less

developed regions by re-estimating equation (1) with the inclusion of the interactive effect

between EM and HQ1 (HQ2). The model is depicted as follows:

𝐹𝑉!" = 𝑏! + 𝑏!𝐸𝑀! + 𝑏!𝐻𝑄!" + 𝑏!𝐸𝑀!×𝐻𝑄!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝐶𝐹𝑂!" + 𝑏!𝑃𝑃𝐸!" +𝑏!𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" + 𝑏!"𝐵𝐼𝐺!" + 𝑏!!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" + 𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!"𝐷𝑂𝑀!" +𝑒!" (3)

where HQ = HQ1 or HQ2. The key coefficient of interest in equation (3) is b3, which we predict

to be negative and significant. If managers are using the fair value model for manipulative

purposes, then we expect that firms located in less developed regions are more likely to report

fair values for investment properties.

3.3. Post-adoption earnings management Fair values of investment properties are generally determined by appraisers, who are supposed to

be experts in valuation and real estate. However, many appraisers in China lack the expertise and

proper education necessary to conduct such appraisals.17 Hence, there are concerns about the

16 For geographic distance, the distribution is highly skewed because both exchanges are located on the coast and are far away from the inland provinces. Hence, we set an arbitrary number (i.e., 500km) as the threshold. 17 For instance, according to the Statistical Bureau in China in 2009, the number of financial experts who hold the CFA designation in China is less than 1,000, while there are already more than 5,000 CFA Charterholders in the city of Hong Kong alone.

18

quality of appraisals in China, especially in remote areas. Managers may purposely work with an

“easy” appraiser so that they can exert more influence on the appraisal process to manipulate fair

value estimates. A study by Dietrich et al. (2001) shows that property fair value estimates are

sometimes used by U.K. firms to overstate earnings before debt issuance or to smooth reported

earnings. In an emerging market such as China, where regulations are weak and unevenly

applied and where ethical standards are not well developed, it is even more likely that managers

would be more likely to bias their fair values estimates. We specifically examine whether

Chinese managers use fair value estimates to either smooth reported earnings or to beat earnings

benchmark. Trueman and Titman (1988) demonstrate that smoothing activities could reduce the

volatility of earnings and could result in lower estimated bankruptcy costs. We posit that firms

choosing the fair value model for their investment properties would manipulate the fair value

estimates for earnings management purposes, and that this behavior is most serious for firms

located in less developed regions.

H3a: Firms that have adopted fair value reporting for investment properties are more likely to

engage in earnings management post-CAS 3 compared with firm using the cost model.

H3b: Firms located in less developed regions that choose the fair value reporting model for

investment properties are more likely to engage in earnings management post-CAS 3 compared

with other firms.

If the intention of managers is to adopt the fair value model to smooth earnings, then we

expect managers to report unrealized gains and losses from investment properties in a manner

that reduces the volatility of earnings. Hence, when the change in reported net income (excluding

fair value gains and losses) is more positive (negative), we expect the gains and losses from fair

value adjustments to be more negative (positive). We use the following model to investigate the

relation between fair value gains and losses and earnings change for the fair value adopting firms:

𝐹𝑉𝐺𝐿!" = 𝑏! + 𝑏!∆𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠!" + 𝑏!×%𝐼𝑃!" + 𝑒 (4)

19

FVGLit is the fair value gains and losses recognized from the changes in the values of investment

properties, scaled by the beginning-of-year market value of equity. ∆Earningsit is the change in

reported net income exclusive of fair value gains and losses from an investment property’s fair

value changes, scaled by the beginning-of-year market value of equity. %IPit is the percentage of

investment properties relative to total assets. We predict b1 to be negative if managers use the

unrealized fair value gains and losses to smooth earnings.

Another motivation for earnings management is to meet or beat earnings benchmark (e.g.

positive earnings or increase in earnings). With this motivation, we expect firms that have

adopted the fair value model for investment properties to be more likely to have earnings or

change of earnings very close to or slightly above zero (i.e., we do not consider beating analyst

forecasts as these are not available for most of our sample of Chinese firms). We use the

following model to investigate the adopting firms’ behaviors of meeting or beating earnings

benchmark:

𝑆𝑢𝑠𝑝𝑒𝑐𝑡!" = 𝑏! + 𝑏!𝐹𝑉!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝑀𝐵!" + 𝑏!𝐶𝐹𝑂!" + 𝑒!" (5)

𝑆𝑢𝑠𝑝𝑒𝑐𝑡!" is an indicator variable that equals one for firms whose earnings (or change of

earnings) scaled by total assets are between 0 and 0.005, zero otherwise. MBit is the market-to-

book ratio. The other variables are defined in previous models. Prior studies (e.g., Burgstahler

and Dichev 1997; Dechow et al. 2003) have argued that it is likely that firms with earnings (or

change of earnings) just below zero manage their earnings to report income marginally above

zero. We predict 𝑏! to be positive if fair value adopting firms manage earnings to meet or beat

earnings benchmark.

4. Sample selection and descriptive statistics

4.1. Sample selection Our sample is obtained from the China Center for Economic Research (CCER) database over the

period of 2007-2009 since CAS 3 became effective in 2007. We hand-collected from firms’

annual reports the accounting choice for investment property (fair value or cost model), the

change in fair value of investment property, and other related investment property information.

Geographic and demographic macro-data is obtained from China’s statistics yearbook of 2006.

Our sample comprises all listed A-share companies that have investment properties, with the

20

exclusion of financial and IPO firms. This results in 1,563 firm-year observations in our main

sample from 579 sample firms. We find that 22 out of 579 firms have switched to the fair value

model subsequent to the implementation of CAS 3 in China. Consistent with anecdotal evidence

(Chinese CA Network 2011), the percentage of firms choosing the fair value model under CAS 3

is small, only approximately 3.8% in our sample. These fair value firms are distributed across 12

different industries, with slightly higher representations from the retail and real estate industries

(i.e., 4 and 3 firms respectively). Finally, we also collect a sample of 2,592 firm-year

observations over 2001-2005 for these 579 sample firms for the calculation of past discretionary

accruals. Table 1 presents the results of the sample selection process.

[Insert Table 1 here]

4.2. Descriptive statistics Table 2, Panel A, reports descriptive statistics for the variables used in the regression analysis.

We again show that very few firms adopt the fair value model, as FV has a mean of 0.0134 in the

sample. This 1.34% rate is lower than the 3.8% noted above because we only include a fair value

adopting firm once whereas other firms are included for multiple years.18 The measures of

economic development, HQ1 and HQ2, report mean values of 0.52 (range between 0 and 1) and

4.11 (range between 0 and 7) respectively. The sample of firms has a mean size of 22.11

measured in logs, or about RMB4 billion, mean leverage ratio of 0.58 and average cash flows

from operations of 0.035. The mean stock return is -0.21, which is consistent with recent

findings (e.g., Yang and Lim 2009) documenting the poor performance of the Chinese stock

market in light of the financial crisis. Average PPE is 0.23, and 12.6% of the sample

observations report net losses for a given year. BIG4 has a mean of only 0.072, indicating the Big

4 auditors are not used by most firms in China. This low percentage of Big 4 auditor use is also

documented by Gul et al. (2010). The corporate governance variables MANAGER, CHAIR_CEO

and DOM have average values of 0.016, 0.14 and 0.45 respectively.

18 Similar to IAS 40, CAS 3 imposes strict restriction on firms switching back from the fair value model to the cost model. As a consequence, we find that none of our sample firms switched back after they adopt the fair value model. To avoid counting the fair value adopting decision of the same firm multiple times over 2007-2009, we only include observations of the fair value adopting firms in the year when they first adopt the fair value model in the panel regression. Hence, FV has mean of 1.34% while the average percentage of adoption is 3.8% for the sample firms.

21


In Table 2, Panel B, we report Pearson correlation coefficients between the regression variables.

We find that FV and EM are significantly positively correlated. We also find that fair value

adopting firms have higher leverage ratios and lower PPE. Consistent with traditional earnings

management literature, EM is higher in smaller firms, firms with lower cash flow from operating

activities, loss firms, and firms without Big 4 auditors. One correlation is noteworthy: HQ1, the

indicator variable for regional economy based on whether a company’s headquarter is located in

one of the three most developed areas in China, is correlated with regional index HQ2 based on

seven economic factors with a correlation coefficient of 0.91. We use both of these measures in

the main regression analysis.

5. Empirical results

5.1. Univariate tests To begin our analysis, we first compare the level of earnings management activity pre-CAS 3 of

firms that later opted for the fair value model vs. those retaining the cost model. The first column

of Table 3 shows that fair value firms reported somewhat higher discretionary accruals, although

the difference is not statistically significant (t = 1.42). Next, we partition the sample observations

based on whether HQ1 equals 1 or 0 (i.e., whether a firm’s headquarter is located in a developed

region or not). The results show that for firms headquartered in developed regions (column 2),

past discretionary accruals are lower for those that later chose fair values for investment

properties, compared to those retaining the cost model (t = 3.65). In contrast, for firms located in

less developed regions (column 3), discretionary accruals are higher for those choosing the fair

value model than those choosing the cost model (t = 2.48). These preliminary results suggest that

past earnings quality is a predictor of firms’ choice of reporting for investment properties, and

firm location is an important mediating factor.


22

5.2. Multivariate logistic regressions For the formal test of H1, we estimate equation (2) using logistic regression. In all regressions,

we control for potential cross-correlations within firms by reporting robust standard errors

clustered at the firm-level. In the first column of Table 4, we report the logistic regression

estimation results for equation (2). The results show that the coefficient for past earnings

management (EM) is significantly positive (p = 0.049), consistent with our hypothesis that past

earnings management behavior is predictive of the likelihood of a firm choosing the fair value

model for investment properties. Economically, this coefficient estimate is small but significant,

as it translates into a marginal effect of 4% of EM on the likelihood of fair value adoption.

In the second and third columns of Table 4, we report results of estimating equation (3),

which examines H2—the effect of location on the association between fair value reporting choice

and past earnings management. Column 2 uses indicator variable HQ1 as the measure of location.

The results show that the coefficient of EM remains positive and becomes more significant as

compared to results in Column 1 discussed above. This coefficient is relevant to firms for which

HQ1 = 0, meaning that for firms in less developed regions, the choice of fair value accounting

associates with past earnings management. The coefficient on HQ1 is positive and significant,

implying that firms in more developed regions are more likely to choose fair value accounting

for investment properties, consistent with the requirements of CAS 3 (i.e., that firms are able to

support the fair value estimates with information from an active estate market). The coefficient

on the interactive term EM×HQ1 is negative and strongly significant (p = 0.0021), showing that

for firms in developed (less developed) regions, the likelihood of choosing the fair value model

is negatively (positively) associated with prior earnings management. Economically, the

marginal effect of EM on the likelihood of fair value adoption increases from 4% to 10% when

firms are located in less developed regions. We obtain similar results in Column 3 using HQ2 as

the measure of location.

Of the control variables, leverage ratio (LEV) is significantly positive while PPE ratio

(PPE) is significantly negative, which implies risky firms and firms with lower tangibility are

more likely to choose the fair value model. We also find significant and negative coefficients for

managerial shareholdings (MANAGER).


23

We next examine whether firms that adopt fair value reporting for investment properties

subsequently exhibit evidence of stronger earnings management. In Table 5, we report the results

of estimating equation (4) using 51 firm-year observations with reported unrealized fair value

gains and losses on investment property (FVGL). Column (1) shows that FVGL is negatively

associated with the change in non-FVGL earnings, consistent with earnings smoothing. Columns

(2) and (3) examine the impact of location using HQ1 and HQ2 as alternative proxies. While we

do not find any significance for the interaction term of ∆Earnings and HQ1, the interactive

coefficient is about 80% of the magnitude of the main effect, such that sum of the coefficients on

∆Earnings and ∆Earnings×HQ1 becomes insignificant (F = 0.19, p = 0.6647), indicating that

location in one of the three developed regions of China fully mitigates the smoothing effect. The

results for HQ2 are similar except that the coefficient on the interaction term between ∆Earnings

and HQ2 is significantly positive (p = 0.0023), again indicating that earnings manipulation

behavior is mitigated when firm headquarters are located in more developed regions. The

interpretation is not as simple as for HQ1 because HQ2 ranges from 0 to 7. Applying the inter-

quartile range of 7 yields an interactive effect of 7 × 0.07841 = 0.54887, which is 127% of the

magnitude of the main effect of -0.4328. Therefore, the evidence shows that location in a more

developed region fully mitigates the tendency to use of fair value gains and losses to smooth

earnings.19


A second common motivation for earnings management is to meet earnings benchmarks. In

Table 6, we report the results of testing whether fair value adopting firms are more likely to meet

or beat earnings benchmarks than firms using the cost model. The analysis uses 7,673 firm-

quarters with data for earnings and earnings changes in 2007 to 2009. Column 1 shows the

logistic regression results of meeting or beating the zero earnings level benchmark. We find the

coefficient for the fair value indicator FV is positive and significant (p = 0.043) indicating that

firms that use the fair value model for investment properties are more likely to meet or beat the

19 Tangentially, we note that 43 of the 51 instances of fair value adjustments are gains while only 8 are losses.

24

zero earnings threshold. Furthermore, limiting the sample to exclude the 48 firm-quarters with

fair value losses (see Column 2), the effect becomes stronger as would be expected (p = 0.0239).


Table 7 shows the results of the analysis of meeting or beating the benchmark of zero

change in earnings. The positive and statistically significant coefficients of the fair value

indicator (p = 0.0001) again implies that fair value adopting firms are able to meet or beat the

zero earnings changes threshold more often than firms using the cost model. Similar to Table 6,

the effect become stronger when we exclude observations with fair value losses.


Finally, we provide in Figures 1 and 2 some graphical evidence of the frequency of fair

value adopting firms beating or meeting earnings or earnings change benchmarks. Figure 1

presents distribution of quarterly earnings scaled by total assets for fair value model firms and

cost model firms separately. We group earnings into bin widths of 0.005 and plot the

observations from -0.12 to 0.12. Each interval is defined to include its lower boundary and

exclude its upper boundary. We observe that the distribution for fair value firms shows a slightly

more obvious kink to the left of the zero earnings benchmark. A similar, albeit less apparent,

pattern is observed in Figure 2 which depicts distribution of change of quarterly earnings scaled

by total assets.

6. Robustness tests

6.1. The impact of investment property location In the analysis above, we show that the location of firm headquarters play an important role in

financial reporting decision-making. This result could be due to two reasons. It is possible that

firms in more developed regions are subject to stronger investor monitoring, which generally

constrains managerial opportunism. Alternatively, investment properties in these regions allow

less potential for appraisal misstatement. To distinguish these two explanations, we hand-collect

detailed data of investment property locations from firms that voluntarily disclose this

information. After checking all sample firms’ annual reports during 2007 to 2009, we obtain a

25

sample of 448 firm-year observations. For this subsample, we find that about 74% have the

majority of their investment properties (i.e., greater than 75% in terms of total number of

investment properties) located in the same region as their firms’ headquarters. Since firms have

multiple investment properties that may be located in different regions, we build a new index,

IPLOC, to proxy for the average economic development level of the firms’ investment property

locations. If firms have all their investment properties in the region of the headquarter, then

IPLOC equals to HQ2. If firms have investment properties in different regions, then IPLOC

equals to the mean of the HQ2 score from each of those regions.20

Next, we substitute IPLOC as our new proxy of location in equation (3) to examine the

impact of investment property location on fair value choice. The first column of Table 8, Panel A,

reports the regression results. The coefficient of interaction between IPLOC and EM is

significantly negative (p = 0.028), implying the location of investment properties does have an

effect on fair value choice. In order to examine whether investment property locations or

headquarter locations are more significant factors in determining fair value reporting decision,

we run a comparison and replicate the same analysis on equation (3), using HQ1 and HQ2 as the

location proxies on the same subsample of 448 observations. Comparing the three columns in

Table 8, Panel A, the Vuong statistics show Model 1 is indistinguishable from Model 2 using

HQ1, Model 3 is statistically significantly better than Model 1 using HQ2. We further compare

the effect of IPLOC and HQ which we include both variables (and their interaction terms) in the

same specification. In the first column of Table 8, Panel B, we show that the interaction term of

IPLOC and EM is significant but not the interaction term of HQ and EM. In the second column,

both interaction terms are significant. The mixed findings seem to imply that investor monitoring

and real estate market efficiency both impact the managerial choice with regards to fair value

reporting for investment properties.


20 For example, if firm A has investment properties in region 1, region 2 and region3, which have HQ2 scores of 2, 6, and 7 respectively, then the IP score for firm A is equal to 5. Ideally, we would want to construct a weighted index based on proportional investment in different regions. Unfortunately, while these firms disclose each of their investment properties in terms of location, not every firm provides detailed breakdown of the values of the individual investment properties.

26

6.2. Alternative proxies for earnings management In our main regression analysis, earnings management is measured by firms’ past five years’

average discretionary accruals using the Dechow and Dichev (2002) model. Although Jones et al.

(2010) compares several accrual models and finds that the model of Dechow and Dichev (2002)

outperforms other models in its ability to detect actual cases of fraudulent and restated earnings;

we check the robustness of our results by using the Modified Jones’ model to calculate

discretionary accruals. We follow the methodology in Larcker and Richardson (2004) to

calculate discretionary accruals for our variable EM. Table 9 reports the results. The likelihood

of choosing the fair value model for investment property remains positively associated with the

mean absolute discretionary accruals (p = 0.007). When we compare discretionary accruals with

our location factors, HQ1 and HQ2, we again find strong results similar to those reported in

Table 4, re-affirming the positive association between earnings management and the probability

of using the fair value model, especially in less developed regions.

[Insert Table 9]

7. Conclusion The issue of reliability has been the main concern of fair value reporting, and fair value estimates

for nonfinancial assets are especially susceptible to managerial manipulation. Given the

opportunity, managers could abuse the flexibility allowed in the fair value model to report

unrealistic and unreliable fair value estimates. In this study, we hypothesize that firms in the

emerging market would adopt the fair value model as an earnings management tool. We utilize

the emerging market setting in China and empirically examine the following questions: 1) How

do firms make use of fair value reporting? 2) What is the role of location in the choice of fair

values for investment properties? We find strong and robust evidence that the fair value model

for investment properties in China is chosen more often by firms that have exhibited more

earnings management activity in the past, and the likelihood of these firms choosing the fair

value model increases when the firms’ headquarters or investment properties are located in less

developed regions. We also find that these fair value adopting firms engage in earnings

smoothing using the unrealized gains and losses from investment properties, and they are also

more likely to meet or beat certain earnings benchmarks (zero earnings and zero earnings

27

change). Indeed, these results indicate that the new accounting standards on fair value reporting

for investment properties (i.e., CAS 3 or, more generally, IAS 40) comes with substantial

concerns when implemented in an emerging market such as China.

Our study bears important policy implications for regulators in emerging economies

when many of these countries are considering or are in the process of converging their local

standards to IFRS. We find evidence that China, the largest emerging market in the world, does

not seem to implement the IFRS well (at least in the accounting of investment properties).

Contrary to the belief that the adoption of IFRS would improve financial reporting quality of a

country, we show that introducing the IFRS-equivalent standards and allowing fair value

reporting for investment properties in China in fact encourages earnings management activities.

Our findings imply fair value reporting may improve the relevance of financial information in

most cases, but fair values may not be superior to historical costs when the concern of reliability

outweighs the benefit of providing relevant information in countries with less efficient market

and lower investor monitoring.

Lastly, our study contributes to the literature by showing that location can affect

accounting-related corporate decision making. Of course, the caveat of our study is that we

examine the financial reporting choice of investment properties, and the valuation of investment

properties is essentially location-driven. Nonetheless, the impact of location is often overlooked

in the accounting literature, and yet, firm location naturally affects the effectiveness of investor

monitoring and the level of information asymmetry between managers and investors. Our study

highlights the importance of location as a factor in financial reporting decision.

28

Data Appendix FV Indicator variable, equal to 1 if firms choose fair value model; 0 otherwise

EM Square root of mean absolute discretionary accruals

HQ1 Indicator variable, equal to 1 if firms are located at developed regions; 0 otherwise

HQ2 Index from 0 to 7 representing the development level of a region

FVGL Unrealized fair value gains and losses

△Earnings Change in reported net income exclusive of unrealized fair value gains and losses from

investment properties

SIZE Natural logarithm of firm’s equity market capitalization

LEV Total liabilities divided by firm’s equity market capitalization

MB Market value of equity over book value of equity

CFO Cash flows from operations divided by firm’s equity market capitalization

PPE Total property, plant and equipment divided by firm’s total assets

%IP Value of investment properties divided by total assets

BIG4 Indicator variable, equal to one if firms hire a BIG4 firm as auditors; 0 otherwise

RETURN Total stock return

LOSS Indicator variable, equal to one if firms have negative net income year; 0 otherwise

MANAGER Indicator variable, equal to one if managers’ total shareholding is equal to or larger

than 5%; 0 otherwise

CHAIR_CEO Indicator variable, equal to one if CEO is also the chair of the board; 0 otherwise

DOM Indicator variable, equal to one if firms’ largest five shareholders’ total shareholdings

are equal or larger than 5%; 0 otherwise

29

References

Aboody, D., M. Barth, and R. Kasznik. 1999. Revaluations of fixed assets and future firm performance: Evidence from the UK. Journal of Accounting and Economics 26 (1-3): 149-178.

Audretsch, D., and M.P. Feldman. 2004. Knowledge spillovers and the geography of innovation. Handbook of Urban and Regional Economics 4: 2713-2739.

Ayers, B., S. Ramalingegowd, and E. Yeung. 2011. Hometown advantage: the effects of monitoring institutions location on financial reporting discretion. Journal of Accounting and Economics 52 (1): 41-61.

Baik, B., J.K. Kang and J.M. Kim. 2010. Local institutional investors, information asymmetries, and equity returns. Journal of Financial Economics 97: 81-106.

Ball, R., A. Robin, and J.S. Wu. 2003. Incentives versus standards: properties of accounting income in four East Asian countries. Journal of Accounting and Economics 36: 235-270.

Barth, M. 1994. Fair value accounting: evidence from investment securities and the market valuation of bank. The Accounting Review 69 (1): 1-25.

Barth, M., and G. Clinch. 1998. Revalued financial, tangible, and intangible assets: associations with share prices and non-market based value estimates. Journal of Accounting Research 36 (Supplement): 199-233.

Blue Book of Chinese Regional Development. 2006. The Institute of Social Science of The People's Republic of China.

Burgstahler, D., and I. Dichev. 1997. Earnings management to avoid earnings decreases and losses. Journal of Accounting and Economics 24: 99-126.

Cai, Y., and X. Tian. 2009. Firm location and takeover likelihood. Working paper, Indiana University.

Cairns, D., D. Massouid, R. Taplin, and A. Tarca. 2011. IFRS fair value measurement and accounting policy choice in the United Kingdom and Australia. The British Accounting Review 43 (1): 1-21.

Chinese CA Network. 2011. 浅谈推动我国投资性房地产公允价值估值技术应用的几点思考. Available at: http://www.CAnet.com.cn/.

Christensen, H., and V. Nikolaev. 2009. Who uses fair value accounting for non-financial assets after IFRS adoption? Working paper, University of Chicago.

Christoffersen, S. and S. Sarkissian. 2009. City size and fund performance. Journal of Financial Economics 92 (2): 252-275.

Coval, J., and T.J. Moskowitz. 1999. Home bias at home: local equity preference in domestic portfolios. Journal of Finance 54: 2045-2073.

Coval, J., and T.J. Moskowitz. 2001. The geography of investment: informed trading and asset prices. Journal of Political Economy 109: 811-841.

Cumming, D.J. 2006. The determinants of venture capital portfolio size: empirical evidence. Journal of Business 79: 1083-1126.

Danbolt, T., and W. Rees. 2008. An experiment in fair value accounting: UK investment vehicles. European Accounting Review 17 (2): 271-303.

Dechow, P., and I. Dichev. 2002. The quality of accruals and earnings: the role of accrual estimation errors. The Accounting Review 77: 35-59.

Dechow, P., S.A. Richardson, and I. Tuna. 2003. Why are earnings kinky? Review of Accounting Studies 8: 355-384.

30

DeFond, M., and J. Jiambalvo. 1994. Debt covenant violation and manipulation of accruals. Journal of Accounting and Economics 17 (1-2): 145-176.

Degeorge, F., J. Patel, and R. Zeckhauser. 1999. Earnings management to exceed thresholds. Journal of Business 72: 1-33.

Dietrich, J.R., M.S. Harris, and K.A. III Muller. 2001. The reliability of investment property fair value estimates. Journal of Accounting and Economics 30 (2): 125-158.

Easton, P., P. Eddey, and T. Harris. 1993. An investigation of revaluations of tangible long-lived assets. Journal of Accounting Research 31: 1-38.

Edelstein, R., S. Fortin, and D. Tsang. 2012. An international exploration of financial reporting practices in the real estate industry. International Real Estate Review 15 (3): 253-281.

Fields, T., T. Lys, and L. Vincent. 2001. Empirical research on accounting choice. Journal of Accounting and Economics 31 (1-3): 255-307.

Francis, B., H. Iftekhar, and M. Waisman. 2008. Does geography matter to bondholders? Working paper, Rennsealer Polytechnic Institute.

Fritsch, M. 2003. Does R&D-cooperation behavior differ between regions? Industry &Innovation 10 (1): 25-39.

Glaeser, E., and D. Mare. 2001. Cities and skills. Journal of Labor Economics 19: 316–342. Glaeser, E., J. Scheinkman, and A. Schelifer. 1995. Economic growth in a cross-section of cities.

Journal of Monetary Economics 36:117–134. Global Real Estate Transparency Index 2010 report. Jones Lang LaSalle. Graham, J., C. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial

reporting. Journal of Accounting and Economics 40 (1-3): 3-73. Guidry, F., A.J. Leone, and S. Rock. 1999. Earnings-based bonus plans and earnings

management by business-unit managers. Journal of Accounting and Economics 23 (1-3): 113-142.

Gul, F.A., J.-B. Kim, and A.A. Qiu. 2010. Ownership concentration, foreign shareholding, audit quality, and stock price synchronicity: Evidence from China. Journal of Financial Economics 95 (3): 425-442.

Gupta, A.K., and H.J. Sapienza. 1992. Determinants of capital firms' preferences regarding the industry diversity and geographic scope of their investments. Journal of Business Venturing 7: 347-362.

Healy, P. 1985. The impact of bonus schemes on the selection of accounting principles. Journal of Accounting and Economics 7: 85-107.

Herrmann, D., and S. Saudagaran. 2006. The quality of fair value measures for property, plant and equipment. Accounting Forum 30 (1): 43-59.

Hitz, J.M. 2007. The decision usefulness of fair value accounting --- a theoretical perspective. European Accounting Review 16 (2): 323-362.

Holthausen, R., D. Larcker, and R. Sloan. 1995. Annual bonus schemes and the manipulation of earnings. Journal of Accounting and Economics 19 (1): 29-74.

Ivkovic, Z., and S. Weisbenner. 2005. Local does as local is: information content of the geography of individual investors’ common stock investments. Journal of Finance 60: 267-306.

John, K., A. Knyazeva, and D. Knyazeva. 2008. Do shareholders care about geography? Working paper, New York University.

31

Jones, K., G. Krishnan, and K. Melendrez. 2010. Do models of discretionary accruals detect actual cases of fraudulent and restated earnings? An empirical analysis. Contemporary Accounting Research 25 (2): 499-531.

Kasznik, R. 1999. On the association between voluntary disclosure and earnings management. Journal of Accounting Research 37 (1): 57-81.

Larcker, D.F., and S.A. Richardson. 2004. Fees paid to audit firms, accrual choices, and corporate governance. Journal of Accounting Research 42 (3): 625-656.

Leuz, C., and P. Wysocki. 2008. Economic consequences of financial reporting and disclosure regulation: a review and suggestions for future research. Working paper, University of Chicago.

Leuz, C., D. Nanda, and P. Wysocki. 2003. Earnings management and investor protection: an international comparison. Journal of Financial Economics 69 (3): 505-527.

Li, J., C.W. Lee, and X. Chen. 2008. Government assisted earnings management in China. Journal of Accounting and Public Policy 27 (3): 262-274.

Lo, K. 2008. Earnings management and earnings quality. Journal of Accounting and Economics 45: 350-357.

Loughran, T., and P. Schulz. 2005. Liquidity: Urban versus rural firms. Journal of Financial Economics 78 (2): 341-374.

Loughran, T., and P. Schulz. 2006. Asymmetric information, firm location, and equity issuance. Working paper, University of Notre Dame.

Man, J.Y. 2012. China’s property tax reform progress and challenges. Land Lines. Marquardt, C.A., and C.I. Wiedman. 2004. How are earnings managed? An examination of

specific accruals. Contemporary Accounting Research 21 (2): 461-491. Muller, K., R. Riedl, and T. Sellhorn. 2011. Mandatory fair value accounting and information

asymmetry: evidence from the European real estate industry. Management Science 57 (6) 1138-1153.

Nelson, K. 1996. Fair value accounting for commercial banks: an empirical analysis of SFAS No. 107. The Accounting Review 71 (2): 161-182.

Norton, E. and B. Tenenbaum. 1993. Specialization versus diversification as a venture capital investment strategy. Journal of Business Venturing 8 (5): 431-442.

Perry, S.E., and T.H. Williams. 1994. Earnings management preceding management buyout offers. Journal of Accounting and Economics 18 (2): 157-179.

Quagli, A., and F. Avallone. 2010. Fair value or cost model? Drivers of choice for IAS 40 in real estate industry. European Accounting Review 19 (3): 461-493.

Rauch, J. 1993. Productivity gains and geographic concentration of human capital: Evidence from the cities. Journal of Urban Economics 34: 380–400.

Sloan, R.G. 1999. Evaluating the reliability of current value estimates. Journal of Accounting and Economics 26 (1): 193-200.

Sohu Business. 2008. Investment Property Fair Value Appraisal got a Cold Shoulder. Available at: http://www.business.sohu.com/20080110.

Song, C.J., W. Thomas, and H. Yi. 2010. Value relevance of FAS No.157 fair value hierarchy information and the impact of corporate governance mechanisms. The Accounting Review 85 (4): 1375-1410.

Sorenson, O., and T.E. Stuart. 2001. Syndication networks and the spatial distribution of venture capital financing. American Journal of Sociology 106: 1546-1588.

Tao, X.N. 2008. The distribution map of Chinese stock investors. Financial Weekly.

32

Trueman, B., and S. Titman. 1988. An explanation for accounting income smoothing. Journal of Accounting Research 26: 127-139.

Urcan, O. 2007. Geographical location and corporate disclosures. Working paper, London Business School.

Wang, H., and K. Wang. 2012. What is unique about Chinese real estate markets? Journal of Real Estate Research 34 (3): 275-290.

Watts, R.L., and J.L. Zimmerman. 1978. Towards a positive theory of the determination of accounting standards. Accounting Review 53 (1): 112-134.

Yang, M., and T.S. Lim. 2009. Why did Chinese stock market perform so badly in 2008? EAI Background Brief No. 427.

33

Table 1 Sample selection Initial sample of firm-year observations reported in CCER database for year 2007-2009 6340

Less: firms without investment property (2940)

Less: Financial industries (1395)

Less: IPOs (112)

Less: firms with no financial records before 2006 (330)

Final sample of firm-year observations 1563

Table 2

Panel A: Descriptive statistics

Variables N Mean Median Std Dev Min 1st quartile 3rd quartile Max FV 1563 0.0134 0.0000 0.1152 0.0000 0.0000 0.0000 1.0000 EM 1563 0.2236 0.2004 0.1122 0.0354 0.1479 0.2755 1.1856 HQ1 1563 0.5237 1.0000 0.4996 0.0000 0.0000 1.0000 1.0000 HQ2 1563 4.1118 6.0000 3.0127 0.0000 0.0000 7.0000 7.0000 SIZE 1563 22.1190 21.9905 1.0632 18.6002 21.4027 22.7208 26.4449 LEV 1563 0.5862 0.3741 0.6420 0.0016 0.1930 0.7248 6.9409 CFO 1563 0.0352 0.0279 0.1127 -0.7392 -0.0006 0.0740 0.8007 PPE 1563 0.2299 0.2041 0.1761 0.0000 0.0874 0.3310 0.8742 RETURN 1563 -0.2148 0.5144 1.2781 -6.4859 -1.1749 0.6350 0.9289 LOSS 1563 0.1257 0.0000 0.3316 0.0000 0.0000 0.0000 1.0000 BIG4 1563 0.0720 0.0000 0.2586 0.0000 0.0000 0.0000 1.0000 MANAGER 1563 0.0158 0.0000 0.1247 0.0000 0.0000 0.0000 1.0000 CHAIR_CEO 1563 0.1403 0.0000 0.3474 0.0000 0.0000 0.0000 1.0000 DOM 1563 0.4510 0.0000 0.4978 0.0000 0.0000 1.0000 1.0000 See Date Appendix for variable definitions.

34

Table 2

Panel B: Pearson Correlations

*

P<0.10 **P<0.05 ***P<0.01. See Date Appendix for variable definitions.

FV

EM

HQ1

HQ2

SIZE

LEV

CFO

PPE

RETURN

LOSS

BIG4

MAN

AGER

CHAIR_CEO

DOM

FV 1

EM 0.0412* 1

HQ1 0.0021 -‐0.0295 1

HQ2 -‐0.0083 -‐0.0324 0.9135*** 1

SIZE 0.0075 -‐0.1578*** 0.0837*** 0.09*** 1

LEV 0.0538** -‐0.0208 -‐0.0388 -‐0.0448* -‐0.1652*** 1

CFO -‐0.0363 -‐0.073*** -‐0.0512** -‐0.0578** 0.0604** 0.1433*** 1

PPE -‐0.0638** -‐0.1098*** -‐0.2431*** -‐0.2742*** 0.0485* 0.0202 0.2459*** 1

RETURN 0.008 0.0233 -‐0.004 0.0024 0.3484*** -‐0.4284*** 0.0348 -‐0.0026 1

LOSS -‐0.0274 0.075*** -‐0.0084 -‐0.0172 -‐0.1485*** 0.0305 -‐0.0329 0.0403 0.0652*** 1

BIG4 -‐0.0053 -‐0.0883*** 0.0993*** 0.104*** 0.3244*** 0.1225*** 0.1372*** 0.0961*** -‐0.0361 -‐0.0098 1

MANAGER -‐0.0182 0.0256 0.0802*** 0.0811*** -‐0.0063 -‐0.0623** -‐0.0097 -‐0.0383 0.0212 -‐0.0327 -‐0.0353 1

CHAIR_CEO 0.0066 0.0471 0.0922 0.0937 -‐0.1056 -‐0.0868 -‐0.0069 -‐0.0056 0.0085 0.0305 -‐0.0699 0.0805 1

DOM -‐0.0129 0.0055 0.0383 0.0396 0.2699*** 0.0376 -‐0.0019 0.0009 -‐0.0224 -‐0.0641*** 0.1452*** -‐0.0028 -‐0.0993*** 1

35

Table 3 Univariate test of differences in absolute discretionary accruals

ALL

FIRMS n

HQ1 = 1 (firms in

developed regions) n

HQ1 = 0 (firms in less

developed regions) n

FV = 1 (fair value firms)

0.0835 101 0.043 45 0.116 56

FV = 0 (cost model firms)

0.0649 2491 0.0676 1299 0.062 1192

Difference

0.0186

2592

-0.024

0.054

t-value 1.42 3.65*** 2.48** Number of observations

1248 1344

*P<0.10 **P<0.05 ***P<0.01

Discretionary accruals are calculated using Dechow and Dichev (2002) model:

ACCRit=b0+b1CFOi,t-1+b2CFOi,t+b3CFOi,t+1+b4∆REVit+b5PPEit +eit (1)

ACCR is calculated as earnings less operating cash flows. The CFOt-1 and CFOt and CFOt+1 are operating cash flows at year t-1, t, and t+1

respectively. ∆REVit is the difference in total revenue between year t and year t-1. And PPEit is the gross value of property, plant and

equipment. All variables are scaled by total assets at year t. The sample period spans the years 2001- 2005.

36

Table 4 Logistic regression analysis of fair value choice and earnings management

Variables Model(1) Model 2

HQ = HQ1 Model 3

HQ = HQ2 Intercept -6.664

(0.1613) -7.2276

(0.1568) -7.2911

(0.1565) EM 2.1929**

(0.0490) 6.855*** (0.0036)

7.5125*** (0.005)

HQ

1.7785** (0.0361)

0.252* (0.0895)

EM× HQ

-7.8209*** (0.0021)

-1.2291*** (0.0036)

SIZE 0.0845 (0.6852)

0.0521 (0.8132)

0.055 (0.8022)

LEV 0.7081*** (0.0002)

0.7916*** (<.0001)

0.798*** (<.0001)

CFO -0.6557 (0.6471)

-0.7269 (0.6031)

-0.6908 (0.6142)

PPE -3.0672** (0.040)

-2.9776** (0.0482)

-3.2299** (0.0338)

RETURN -0.2567 (0.1779)

-0.2247 (0.2566)

-0.2181 (0.2809)

LOSS -0.848 (0.2863)

-0.8884 (0.2687)

-0.8232 (0.3012)

BIG4 -0.1171 (0.8705)

-0.1157 (0.8828)

-0.1149 (0.8863)

MANAGER -12.573*** (<.0001)

-12.4313*** (<.0001)

-12.359*** (<.0001)

CHAIR_CEO 0.2806 (0.5729)

0.3116 (0.5307)

0.3518 (0.4826)

DOM -0.2618 (0.4654)

-0.2715 (0.4595)

-0.2906*** (0.4302)

Industry indicators YES YES YES Calendar year indicators YES YES YES Pseudo-R2 0.0218 0.0263 0.0266 Max_rescaled R2 0.1206 0.1485 0.1474 n 1563 1563 1563 The following regressions are estimated for the sample period 2007-2009 using panel data with firm-level clustered standard errors for

the estimation of the p-values.

𝐹𝑉it = 𝑏0 + 𝑏1𝐸𝑀i + 𝑏2𝑆𝐼𝑍𝐸!" + 𝑏3𝐸𝑉!" + 𝑏4𝐶𝐹𝑂!" + 𝑏5𝑃𝑃𝐸!" + 𝑏6𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" + 𝑏!𝐵𝐼𝐺4!" + 𝑏!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" +

𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!!𝐷𝑂𝑀!" + 𝑒!" (2)

𝐹𝑉!" = 𝑏! + 𝑏!𝐸𝑀! + 𝑏!𝐻𝑄!" + 𝑏!𝐸𝑀!×𝐻𝑄!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝐶𝐹𝑂!" + 𝑏!𝑃𝑃𝐸!" + 𝑏!𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" + 𝑏!" ∗

𝐵𝐼𝐺!" + 𝑏!!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" + 𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!"𝐷𝑂𝑀!" + 𝑒 (3)

*P<0.10 **P<0.05 ***P<0.01; See Date Appendix for variable definitions.

37

Table 5 Post-adoption earning manipulation test: earnings smoothing

Variables Model (1)

Model(2)

HQ = HQ1

Model(3)

HQ = HQ2

Intercept

0.01377**

(0.015)

0.01431**

(0.0398)

0.01572**

(0.045)

△Earnings

-0.254***

(0.0003)

-0.2947***

(0.0002)

-0.4328***

(<.0001)

HQ

-0.0052

(0.6267)

-0.0016

(0.3072)

△Earnings × HQ

0.23304

(0.181)

0.07841***

(0.0023)

%IP

0.01462

(0.1309)

0.01582

(0.1074)

0.01788*

(0.0594)

Adjusted R2 0.225 0.223 0.35

n 51 51 51 The following regression is estimated for the sample period 2007-2009 using panel data of all firms that adopts fair value reporting for

investment property, with firm-level clustered standard errors for the estimation of the p-values.

𝐹𝑉𝐺𝐿!" = 𝑏! + 𝑏!∆𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠!" + 𝑏!×%𝐼𝑃!" + 𝑒!" (4)

*P<0.10 **P<0.05 ***P<0.01; See Data Appendix for variable definitions.

38

Table 6 Post-adoption earning manipulation test: meet or beat zero earnings benchmark

Zero Earnings Threshold

Variables All firms Excluding

firms with FV loss Excluding

firms with FV gain Intercept 1.664*

(0.0739) 1.793* (0.058)

1.9583** (0.0339)

FV 0.3844** (0.0432)

0.4452** (0.0239)

0.2299 (0.4264)

SIZE -0.1439*** (0.0005)

-0.1496*** (0.0004)

-0.1568*** (0.0001)

LEV -0.00752 (0.9422)

-0.0114 (0.9132)

-0.0139 (0.8979)

CFO -4.2988*** (<.0001)

-4.3507*** (<.0001)

-4.2912*** (<.0001)

MB -0.1635*** (<.0001)

-0.1668*** (<.0001)

-0.1624*** (<.0001)

Industry indicators YES YES YES Calendar year indicators YES YES YES

Pseudo R2 0.02 0.0204 0.02 n 7673 7625 7457

The following regression is estimated for the sample period 2007-2009 using panel data with firm-level clustered standard errors for


𝑆𝑢𝑠𝑝𝑒𝑐𝑡!" = 𝑏! + 𝑏!𝐹𝑉!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝑀𝐵!" + 𝑏!𝐶𝐹𝑂!" + 𝑒!" (5) *P<0.10 **P<0.05 ***P<0.01; See Data Appendix for variable definitions.

39

Table 7 Post-adoption earning manipulation test: meet or beat zero earnings change

benchmark

Zero Earnings Change Threshold

Variables All firms Excluding

firms with FV loss Excluding

firms with FV gain Intercept 0.6815

(0.7136) 0.7703

(0.6786) 1.0771

(0.5646)

FV 0.7415*** (0.0001)

0.7963*** (<.0001)

0.4819* (0.0753)

SIZE -0.09 (0.2126)

-0.0943 (0.1927)

-0.1102 (0.1326)

LEV -0.0455 (0.7137)

-0.0506 (0.6827)

-0.0095 (0.9341)

CFO -3.2869** (0.0286)

-3.2455** (0.0297)

-3.2437** (0.0319)

MB -0.1778 (0.2122)

-0.1755 (0.2129)

-0.1756 (0.2150)

Industry indicators YES YES YES Calendar year indicators YES YES YES

Pseudo R2 0.018 0.0184 0.015 n 7673 7625 7457

The following regression is estimated for the sample period 2007-2009 using panel data with firm-level clustered standard errors for


𝑆𝑢𝑠𝑝𝑒𝑐𝑡!" = 𝑏! + 𝑏!𝐹𝑉!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝑀𝐵!" + 𝑏!𝐶𝐹𝑂!" + 𝑒!" (5) *P<0.10 **P<0.05 ***P<0.01; See Data Appendix for variable definitions.

40

Table 8 Panel A: Logistic regression analysis of fair value choice and investment property locations (Comparing the effect of IPLOC and HQ)

Variables Model(1) Model(2)

HQ = HQ1 Model(3)


(0.9327) -5.1946

(0.5761) -6.2104

(0.5123) EM 8.9888**

(0.0397) 7.4917** (0.0423)

12.7717** (0.0231)

IPLOC 0.2303 (0.3407)

EM×IPLOC -1.5074** (0.0284)

HQ

1.2129 (0.3790)

0.3811 (0.2025)

EM×HQ

-8.7081** (0.0312)

-2.1603** (0.0179)

SIZE -0.2312 (0.5491)

-0.0382 (0.9247)

-0.0545 (0.8930)

LEV 0.4078 (0.3576)

0.5206 (0.2437)

0.5444 (0.2506)

CFO -1.5437 (0.4156)

-2.0397 (0.2957)

-1.7538 (0.3782)

PPE -3.9421 (0.1473)

-3.2706 (0.1917)

-3.3410 (0.1978)

RETURN -0.2849 (0.3404)

-0.3683 (0.3634)

-0.4572 (0.1507)

LOSS -0.8682 (0.3795)

-0.7452 (0.3710)

-0.6160 (0.5065)

BIG4 -11.2540*** (<.0001)

-11.3768*** (<.0001)

-11.2020 (<.0001)

MANAGER -11.2839*** (<.0001)

-12.2004*** (<.0001)

-11.9945*** (<.0001)

CHAIR_CEO -0.2505 (0.8201)

0.1916 (0.8246)

0.2511 (0.7707)

DOM 0.0640 (0.9128)

0.0716 (0.9034)

-0.0368 (0.9527)

Industry fixed effect YES YES YES Calendar fixed effect YES YES YES Pseudo-R2 0.0588 0.0578 0.0631 Max_rescaled R2 0.2120 0.2084 0.2276 N 448 448 448 Vuong Test Model(1) Model(2) Model(3) (1) vs. (2) (1) vs. (3) Log Likelihood -2092 -2105 -2126 Z = 1.1165

(P = 0.2642) Z = 2.8081***

(P = 0.005)

41

Table 8 Panel B: Logistic regression analysis of fair value choice and investment property locations (Including both IPLOC and HQ)

Variables Model(2)

HQ = HQ1 Model(3)


(0.5460) -7.3273

(0.5064) EM 13.3809**

(0.0137) 16.0714**

(0.0150) IPLOC 0.3539 0.3056 (0.2529) (0.2549) EM×IPLOC -1.6518** -1.3419* (0.0363) (0.0517) HQ 0.7538

(0.6195) 0.2387

(0.3656) EM×HQ -6.8929

(0.1329) -1.5523* (0.0687)

SIZE -0.0357 (0.9375)

-0.0444 (0.9211)

LEV 0.4797 (0.3159)

0.4912 (0.3004)

CFO -1.8322 (0.3815)

-1.6946 (0.4078)

PPE -3.4213 (0.2071)

-3.34707 (0.2057)

RETURN -0.4288* (0.0850)

-0.4861** (0.0480)

LOSS -0.8146 (0.3129)

-0.7002 (0.4416)

BIG4 -11.3462*** (<.0001)

-11.2339*** (<.0001)

MANAGER -12.2004*** (<.0001)

-11.7650*** (<.0001)

CHAIR_CEO 0.0836 (0.9279)

0.1474 (0.8729)

DOM 0.1104 (0.8520)

0.0138 (0.9825)

Industry fixed effect YES YES Calendar fixed effect YES YES Pseudo-R2 0.0674 0.0684 Max_rescaled R2 0.2431 0.2467 N 448 448

The regression is estimated for the sample period 2007-2009 using panel data with firm-level clustered standard errors for the

estimation of the p-values.

𝐹𝑉!" = 𝑏! + 𝑏!𝐸𝑀! + 𝑏!𝐻𝑄!" + 𝑏!𝐸𝑀!×𝐻𝑄!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝐶𝐹𝑂!" + 𝑏!𝑃𝑃𝐸!" + 𝑏!𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" +

𝑏!"𝐵𝐼𝐺4!" + 𝑏!!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" + 𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!"𝐷𝑂𝑀!" + 𝑒!" (3)

Since investment property location is not required to disclose in the financial reports, we hand-collected a sample of 448 firms which

have their investment property locations disclosed. For those having investment properties at a single region, IP is equal to the HQ2

index. For those having investment properties at multiple regions, IPLOC is equal to the mean of HQ2 scores from each of those

regions.


42

Table 9 Logistic regression analysis of fair value choice and earnings management (with alternative measure of discretionary accruals)

Variables Model(1) Model(2)

HQ = HQ1 Model(3)

HQ = HQ2 Intercept -8.0812*

(0.0861) -8.3991* (0.0957)

-7.7771 (0.139)

EM 5.4213*** (0.0068)

10.7702*** (0.0001)

12.5245*** (0.0001)

HQ

2.1472** (0.0246)

0.3601** (0.0287)

EM×HQ

-10.7254*** (0.003)

-1.9221*** (0.0015)

SIZE 0.1193 (0.5534)

0.0774 (0.7192)

0.0388 (0.8637)

LEV 0.754*** (0.0001)

0.8616*** (<.0001)

0.8818*** (<.0001)

CFO -0.6606 (0.6466)

-0.5789 (0.664)

-0.4002 (0.7613)

PPE -2.8772* (0.0502)

-2.9091** (0.0427)

-3.2014** (0.0243)

RETURN -0.2605 (0.1826)

-0.1959 (0.3214)

-0.1877 (0.3585)

LOSS -0.98 (0.2193)

-1.0428 (0.2032)

-0.9639 (0.2325)

BIG4 -0.1742 (0.8101)

-0.0598 (0.9368)

0.00356 (0.9963)

MANAGER -12.4459*** (<.0001)

-12.2996*** (<.0001)

-12.1783*** (<.0001)

CHAIR_CEO 0.2723 (0.5895)

0.294 (0.5577)

0.3213 (0.531)

DOM -0.3087 (0.3871)

-0.3433 (0.3467)

-0.3276 (0.3685)

Industry indicators YES YES YES Calendar year indicators YES YES YES Pseudo-R2 0.0238 0.0276 0.0288 Max_rescaled R2 0.1355 0.1568 0.1638 n 1620 1620 1620 The following regression is estimated for the sample period 2007-2009 using panel data with firm-level clustered standard errors for


𝐹𝑉!" = 𝑏! + 𝑏!𝐸𝑀! + 𝑏!𝐻𝑄!" + 𝑏!𝐸𝑀!×𝐻𝑄!" + 𝑏!𝑆𝐼𝑍𝐸!" + 𝑏!𝐿𝐸𝑉!" + 𝑏!𝐶𝐹𝑂!" + 𝑏!𝑃𝑃𝐸!" + 𝑏!𝑅𝐸𝑇𝑈𝑅𝑁!" + 𝑏!𝐿𝑂𝑆𝑆!" +

𝑏!"𝐵𝐼𝐺!" + 𝑏!!𝑀𝐴𝑁𝐴𝐺𝐸𝑅!" + 𝑏!"𝐶𝐻𝐴𝐼𝑅_𝐶𝐸𝑂!" + 𝑏!"𝐷𝑂𝑀!" + 𝑒!" (3)

Earning management potential, EMi, is calculated using the modified Jones Model based on Larcker and Richardson (2004)

ACCRit=b0+b1(1/ATit-1)+b2(∆REVit -∆ARit )+b3 PPEit +b4BMit+b5CFOit+𝑒!"

Where ∆ARit is the change of accounting receivables; BMit is the book value of common equity over the market value of common

equity; CFOit is operating cash flows. All variables are scaled by total assets at year t-1. The sample period spans the years 2000- 2006


43

Figure 1 Level of Earnings Earnings scaled by firms’ total assets. Each interval is of width 0.005.

Non-FV firms

FV firms

FV=0

-0.120 -0.110 -0.100 -0.090 -0.080 -0.070 -0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120

0

2.5

5.0

7.5

10.0

12.5

15.0

Percent

Earning

FV=1

-0.120 -0.110 -0.100 -0.090 -0.080 -0.070 -0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120

0

5

10

15

20

25

Percent

Earning

44

Figure 2 Change of Earnings Change of earnings scaled by firms’ total assets. Each interval is of width 0.005.

Non-FV firms

FV firms

FV=0

-0.120 -0.110 -0.100 -0.090 -0.080 -0.070 -0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120

0

2.5

5.0

7.5

10.0

12.5

15.0

17.5

Percent

CHE

FV=1

-0.120 -0.110 -0.100 -0.090 -0.080 -0.070 -0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0 0.010 0.020 0.030 0.040 0.050 0.060 0.070 0.080 0.090 0.100 0.110 0.120

0

2.5

5.0

7.5

10.0

12.5

15.0

17.5

20.0

22.5

Percent

CHE

Documents

Earnings management, firm location, and financial reporting choice